bigdata4atm whitepaper - may 2016

8/16/2019 BigData4ATM WhitePaper - May 2016

1/22

Big Data Analytics for Socioeconomic

and Behavioural Research in ATM

State-of-the-art and Future Challenges

May 2016


2/22

Big Data Analytics for Socioeconomic and Behavioural Research in ATM

© BigData4ATM Consortium Page 2 of 22

Contents

EXECUTIVE SUMMARY ............................................................................................................................. 3

1. INTRODUCTION ................................................................................................................................... 4

2. NEW BIG DATA SOURCES FOR THE ANALYSIS OF TRAVEL BEHAVIOUR ................................................... 5

3. SOCIOECONOMIC AND BEHAVIOURAL RESEARCH IN ATM: CHALLENGES AND OPPORTUNITIES ............. 6

3.1 PASSENGER-CENTRIC PERFORMANCE METRICS ................................................................................................6

3.2 AIR TRAFFIC FORECASTS ..............................................................................................................................7

3.3 INTEGRATED OPTIMISATION OF AIRPORT LANDSIDE AND AIRSIDE PROCESSES ........................................................8

3.4 SOCIOECONOMIC IMPACT OF ATM DISRUPTIONS ............................................................................................9

4. THE BIGDATA4ATM PROJECT ............................................................................................................. 10

4.1 PROJECT OBJECTIVES ............................................................................................................................... 10

4.2 APPROACH ............................................................................................................................................ 10

4.3 TARGET OUTCOMES AND EXPECTED IMPACT ................................................................................................ 16

REFERENCES ........................................................................................................................................... 17


3/22



Executive summary

A sound understanding of the behavioural and societal factors that influence transport demand and

supply — including economic, social, demographic and cultural issues — is essential for shaping transport

policies according to the values, needs and expectations of our society. However, research on the

relationship between passengers’ behaviour and air traffic management (ATM) performance is relatively

scarce. There is a lack of understanding of the impact of passengers’ behaviour on ATM, as well as of the

impact of ATM on individual passengers and society at large. Research in these areas has so far been

constrained by the limited availability of behavioural data, typically obtained from cross-sectional (static)

demographic and economic datasets, often consisting of very small samples, and usually complemented with

assumptions about the permanence of behavioural traits over time. The pervasive penetration of modern

information and communication technologies (ICT) and the emergence of big data analytics open new

opportunities to overcome this situation: for the first time, thanks to the proliferation of smart personal

devices and interconnected services, we have large-scale, detailed longitudinal (dynamic) data allowing us to

test hypotheses about travellers’ behaviour.

The overall goal of BigData4ATM is to investigate how different passenger-centric geolocated data can be

analysed and combined with more traditional demographic, economic and air transport databases to

extract relevant information about passengers’ behaviour, and to study how this information can be used

to inform ATM decision making processes. The specific objectives of the project are the following:

1. to develop a set of methodologies and algorithms to acquire, integrate and analyse multiple

distributed sources of non-conventional ICT-based spatio-temporal data — including mobile phone

records, data from indoor geolocation technologies, credit card records and data from Internet social

networks, among others — with the aim of characterise passengers’ behavioural patterns;

2. to develop new theoretical models translating these behavioural patterns into relevant and actionable

indicators for the planning and management of the ATM system;

3. to evaluate the potential applications of the new data sources, data analytics techniques and

theoretical models through a number of case studies relevant for the European ATM system, including

the development of passenger-centric door-to-door delay metrics, the improvement of air traffic

forecasting models, the analysis of intra-airport passenger behaviour and its impact on ATM, and the

assessment of the socioeconomic impact of ATM disruptions.


4/22



1.

Introduction

The paramount goal of the European transport policy, as defined in the European Commission’s 2011 White

Paper on Transport (European Commission, 2011a), is to “establish a system that underpins European

economic progress, enhances competitiveness and offers high quality mobility services while using resources

more efficiently”. Particular emphasis is put on the need for a multimodal, passenger-centric transport

system, able to provide seamless door-to-door travel and facilitate better modal choices. In line with these

objectives, the long-term vision for the European aviation sector outlined in the report ‘Flightpath 2050 -

Europe’s Vision for Aviation’ (European Commission, 2011b) identifies five challenges that aviation will have

to face at the 2050 horizon: Meeting Societal and Market Needs (Challenge 1); Maintaining and Extending

Industrial Leadership (Challenge 2); Protecting the Environment and the Energy Supply (Challenge 3);

Ensuring Safety and Security (Challenge 4); and Prioritising Research, Testing Capabilities and Education

(Challenge 5). The report envisages a passenger-centric air transport system thoroughly integrated with

other transport modes, with the ultimate goal of taking travellers and their baggage from door to door

predictably and efficiently while enhancing passenger experience and rendering the transport system more

resilient against disruptive events. As part of Challenge 1, the Flightpath 2050 report highlights the role of

aviation as an enabler of socioeconomic development and stresses the importance of customer orientation,

drawing attention to the key role of the ATM system in realising this vision and defining five high-level goals:

1. European citizens are able to make informed mobility choices and have affordable access to one

another, taking into account: economy, speed, and tailored level of service.

2. 90% of travellers within Europe are able to complete their journey, door-to-door within 4 hours.

3.

Flights arrive within 1 minute of the planned arrival time regardless of weather conditions. Thetransport system is resilient against disruptive events and is capable of automatically and dynamically

reconfiguring the journey within the network to meet the needs of the traveller if disruption occurs.

4. An air traffic management system is in place that provides a range of services to handle at least 25

million flights a year of all types of vehicles.

5. A coherent ground infrastructure is developed including: airports, vertiports and heliports with the

relevant servicing and connecting facilities, also to other modes.

In contrast with this high-level vision, ATM operations have so far lacked a passenger-oriented perspective,

with performance objectives and decision criteria (e.g., flight prioritisation rules) not necessarily taking into

account the ultimate consequences for the passenger (Cook et al., 2013a). Further research is needed to

provide new insights on the interactions between the ATM system and passengers’ needs, choices and

behaviour. However, current methods used to collect data on passengers’ activities are limited in accuracy

and validity: traditional methods based on observations and surveys present intrinsic limitations (e.g.,

incorrect and imprecise answers, dependence on the availability and willingness to answer of the

interviewed persons, etc.), and they are also expensive and time-consuming; useful data can also be

collected from other sources such as air traffic databases, travel reservation systems or market intelligence

data services (e.g., IATA Passenger Intelligence Services, PaxIS), but these data typically fail to capture

important information, such as door-to-door origin-destination pairs and travel times. The generalised use of

geolocated devices in our daily activities opens new opportunities to collect rich data and overcome many of

the limitations of traditional methods. The very same ICT tools that are enabling new forms of bidirectional

communication with the passenger are also making it possible to gather permanently updated information

on passengers’ activity and mobility patterns with an unprecedented level of detail.


5/22



2. New big data sources for the analysis of travel

behaviour The generalisation of the use of portable communication devices has brought new opportunities for

collecting data on human behaviour, opening the door to a systematic statistical treatment of many social

aspects (Lazer et al., 2009). Some examples include the analysis of the structure of social networks (Liben-

Nowell et al., 2005; Onnela et al., 2007; Grabowicz et al., 2012), human cognitive limitations (Gonçalves et

al., 2011), information diffusion and social contagion (Crane et al., 2008; Lehmann et al., 2012), the role of

social groups (Onnela et al., 2009; Grabowicz et al., 2012), the interaction between social relations and

mobility (Grabowicz et al., 2014; Picornell et al., 2015), language coexistence (Mocanu et al., 2013) or

political movements (Borge-Holthoefer et al., 2011; González-Bailón et al., 2011; Connover et al., 2013).

The analysis of human mobility is one of the questions to which the wealth of new data has notably

contributed (Brockmann et al., 2006; Gonzalez et al., 2008; Balcan et al., 2009; Noulas et al., 2012; Bagrow et

al., 2012; Lenormand et al., 2014; Tizzoni et al., 2014; Jurdak et al., 2015). Statistical characteristics of

mobility patterns have been studied (Brockmann et al., 2006; Gonzalez et al., 2008), finding a heavy-tail

decay in the distribution of displacement lengths across users. Most of the trips are short in everyday

mobility, but some are extraordinarily long. Besides, trips are not directed symmetrically in space, but each

user shows a particular structure (Gonzalez et al., 2008). The duration of stay in each location also shows a

skewed distribution with a few preferred places clearly ranking on the top of the list, typically corresponding

to home and work (Song et al., 2010). Recently, geolocated data have also been used to analyse the

structure of urban areas (Lenormand et al., 2014; Louail et al., 2014; Louail et al., 2015), road and long rangetrain traffic (Lenormand et al., 2014b), land use (Soto et al., 2011; Frías-Martinez et al., 2012; Lenormand et

al., 2015) and mobility between cities (Lenormand et al., 2015b) or countries (Hawelka et al., 2014). The new

ICT technologies are permitting the analysis of human mobility discerning different user profiles in terms of

age, socioeconomic level and place of residence (Lenormand et al. 2015b, Lenormand et al. 2015c). It has

been shown that it is possible to track movements across the full continent and to extract origin-destination

matrices at an unprecedented scale and at a very low cost (Lenormand et al., 2014; Lenormand et al.,

2015b). In addition, the limitations of the different data sources have been explored, finding that at large

scales (larger than 2 km), Twitter and cell phone data provide similar results (Lenormand et al., 2014). All

these results evidence the potential of ICT data, either from online social media, cell phone or credit card

records, to characterise door-to-door mobility, from initial origin to final destination. These data sourcesprovide also access to public content, for instance in Twitter messages. The combination of

content/semantic analysis and geolocation is a powerful resource still to be explored in a systematic way.

The potential of these new data sources for the transport sector is huge, but it also comes with a number of

challenges. We have more data, but not always with explanatory power about the underlying decisions of

individuals. We also have a much higher coverage than with traditional surveys, but this sample size often

comes at the expense of low quality, noisy or biased data. The ability to mine, blend and analyse data from

multiple sources will be of paramount importance to overcome these issues and offer reliable and

comprehensive information.


6/22



3. Socioeconomic and behavioural research in ATM:

challenges and opportunities The profound transformation that the current European ATM system is undergoing goes beyond

technological considerations. With society being one of the stakeholders of the ATM Performance

Partnership (SESAR, 2007; EUROCONTROL, 2010a), users’ requirements need to be better accommodated to

enable the execution of flights as close as possible to the passengers’ intention. Planning for and better

adapting the ATM system to the future requires systematic socioeconomic monitoring to understand the

factors influencing changes in demand and flights patterns (EUROCONTROL, 2009b).

A variety of studies can be found on factors influencing future air traffic on the economic dimension.

However, the societal dimension, essential in the paradigm shift from a flight-centric to a passenger-centricair transport system, has been little explored. Some studies have looked at the perception of air transport

and ATM by the European society and their expectations about it, by means of the compilation of existing

surveys (Kinchin, 2004), the analysis of print media mentions of ATM performance (Mahaud and Arrighi de

Casanova, 2004), and the use of questionnaires and discussion groups (Cook and Tanner, 2005), revealing a

general unawareness of ATM performance and a lack of proper coverage of ATM issues in the press.

Ignorance of the frontiers and areas of action of the different ATM sub-systems is also reflected. Worries and

perceptions vary from country to country, and it is suggested that attitudinal responses are strongly driven

by recent events in a given location (Cook and Tanner, 2005). These studies are however restricted on the

number of people and countries reached and cover only one of the many social aspects affecting air traffic. A

comprehensive assessment of the impact of social and behavioural factors (behavioural attitudes, mobilityneeds, tolerance to waiting times, transport modes preferences, etc.), as well as of the net impact that

SESAR and other ongoing ATM modernisation will have on the passenger, remains to be done. In the

following sections we dig deeper into four application areas addressed by BigData4ATM: (i) passenger-

centric performance metrics, (ii) air traffic forecasts, (iii) airport landside and airside processes integrated

optimisation, and (iv) ATM disruption impact assessment.

3.1 Passenger-centric performance metrics

In addition to monitoring and reporting on the performance of the ATM network in terms of delays, the

Network Manager, through the Central Office for Delay Analysis (CODA), provides a monitoring and analysisfunction for all delay reasons (ATFM, airline, airport, etc.). This enables correlation between airline and

network reported delays, and is used in schedule and turnaround planning, enabling better punctuality.

The SESAR WP-E project POEM (Cook et al., 2012, 2013a, 2013b) has proposed the use of new metrics to

measure the net effect of flight delays on the passenger full trip. The project has shown the importance of

using passenger-centric (rather than flight-centric) metrics for a comprehensive assessment of the ATM

system performance and of the effectiveness of policies aimed to improve passenger rights. Passenger-

centric metrics are built on the knowledge of passengers’ itineraries, which are not easy to get: data sources

are sometimes incomplete, not covering data from markets such as low-fare and charter carriers, and usually

restrict itineraries to the air segment (i.e., they do not contain door-to-door information).

The use of passenger-centric geolocation data opens several opportunities to extend the work started by

POEM by considering door-to-door trips and incorporating the study of the economic consequences of such


7/22



delays (e.g., their impact on future travel decisions). These opportunities include: (i) reconstructing

passenger itineraries beyond the air segment, by building origin-destination matrices from mobile phone

data and geolocation apps; (ii) learning about the economic impact of flights delays on passengerexpenditure patterns, measured from credit card data; and (iii) understanding users’ motivation to travel

again, change transport mode, etc., and acquiring a better knowledge of passengers’ perception of ATM

performance through opinion mining and sentiment analysis. Additionally, these new data sources provide

the capability to capture in almost real time passengers’ behaviour in the presence of special events such as

strikes, flight cancellations, news of political instability in a destination, security policies, etc.

3.2 Air traffic forecasts

Air traffic forecasts are an essential input for ATM stakeholders in order to plan response to future air traffic

needs. The EUROCONTROL’s Statistics and Forecast Service (STATFOR) provides forecasts that are used asdirect inputs into the Network Strategy Plan, the Network Operations Plan and the Network Performance

Plans. These forecasts are also a pre-requisite for the establishment of the unit rates used to calculate the

route and terminal charges. Traffic forecasts are also used by an extensive number of planning departments

of airlines, ANSPs, airports, government authorities, etc. for general planning.

Air traffic forecasts are derived in a range of ways depending on the time and data available and also on the

questions that the forecasts are intended to address. For the purpose of technology renewal, airport design

or new routes assignment, a medium- to long-term perspective is needed, while for contingency actions,

short-term forecast is more relevant.

The most common forecasting models are those based on time series analysis. They rely on the assumptionthat patterns observed historically will continue over time and use trend extrapolation and decomposition

techniques, without seeking to understand the causes behind the observed patterns. Time series models are

not able to explain what might be expected in the future in response to policy changes or alternative future

scenarios, but they are able to respond quickly to changes in demand and can be used as benchmarks

against which more sophisticated models can be compared (Airports Commission, 2013).

Causal methods look for cause-effect relationships between explanatory variables and demand patterns.

Explanatory variables are passenger socioeconomic characteristics, GDP, weather conditions, population,

travel costs, consumer spending, etc. In this group of models we find (i) regression models; (ii) gravity

models, which estimate travel flows between two points (e.g., city pairs, country pairs, etc.) as being

proportional to generation/attraction variables (e.g., the population) and inversely proportional to a

generalised distance or cost between origin and destination. The population considered can be a subset of

the total population, as these models are not behavioural and do not distinguish between different types of

travellers; and (iii) qualitative models (e.g., Delphi method), which rely on the judgement of experts. They

can be used to predict a significant change in historical patterns or in cases where there is limited or no data

available. They are normally used for the assessment of how new technological or other developments

would affect the forecast.

The effect of passengers’ profiles and their relationship with the origin and destination locations are not fully

considered by the aforementioned models. The same origin-destination pairs can be connected by different

routes and by a combination of modes. Other travel modes can benefit certain routes, but at the same timethey can act as competitors in the case of short distance trips. Passenger profile not only affects route, mode

and company election, but also how they are modified by changes in other variables such as price, waiting


8/22



time, etc. Elasticity of air travel demand depends on travel nature: short-haul routes and leisure trips are

more sensitive to travel price than long-haul routes or business travels, while business trips are more

sensitive to GDP changes than leisure trips (IATA, 2008; Airports Commission, 2013).Different models have been developed trying to consider these effects in a consistent manner, in an attempt

to predict demand under technological and socioeconomic changes (ICAO, 2006; Airports Commission, 2013;

OECD ITF, 2014). However, these models require large amounts of specialised data, and often need to be

simplified due to the lack of quality data (OECD ITF, 2014). Historically used data sources such as MIDT / BSP

and PaxIS are insufficient or difficult to get due to legal issues. Additionally, some aspects of passengers

profile, real origin-destination pairs and travel purposes are not reflected by such data, so they have to be

obtained from surveys that are expensive, time consuming, and provide small samples. The large amount of

data produced by the extended use of geolocated devices offers new opportunities to replace and/or

complement these data and get new behavioural insights.

3.3 Integrated optimisation of airport landside and airside processes

Passengers’ behaviour at the terminal and landside has a major impact on flights delays (EUROCONTROL,

2009a). Many companies and airports are analysing and testing different options to reduce bottlenecks and

waiting times at the check-in, security, passport control and gate queues. The modelling of passenger

movement at the airport terminal has traditional made use of aggregated approaches, where the behaviour

of single entities (agents) is represented by aggregated flows. However, aggregated models face important

limitations. The fundamental weakness is that they are not based on a coherent theory of passenger

behaviour, and therefore are not suitable to capture agents’ responses to measures influencing their

behaviour. In an attempt to address these shortcomings, agent-based microsimulation models have been

developed considering agents’ intentions, environmental perception and individual interactions. Schultz and

Fricke (2011) have proposed an individual-based movement model based on a stochastic approach, including

system knowledge about characteristics of handling processes, infrastructure knowledge (navigation,

orientation) and perception and processing of provided information (signage). Other analyses have

additionally incorporated passenger group dynamics (Cheng, 2014), showing that group behaviour has

significant influence on the performance and utilisation of airport services. Different companies are also

developing multi-agent airport terminal simulators, such as CAST Terminal, developed by the Airport

Research Center. CAST Terminal offers a three-dimensional environment combined with an agent-based

model of passenger movement behaviour to identify bottlenecks and solutions to complex operational

problems and planning tasks. The relevance of the subject is proven by the companies involved in the

simulator development, including BAA, Frankfurt Airport, Zurich Airport, EUROCONTROL and Airbus.

However, the practical application of agent-based models is still hindered by the need for detailed data to

specify, calibrate, and validate the models. The collection of the data produced by the extended use of

geolocated mobile devices opens new opportunities to overcome this problem. Thanks to indoor position

systems (iBeacons, Wi-Fi-based and Bluetooth-based positioning systems, etc.), we are now able to track

passenger movements in a more detailed manner. The challenge here is to develop appropriate methods for

merging and analysing these data sources so as to extract consistent and reliable information on passenger

movement patterns, and build new models with a stronger behavioural basis allowing the integrated

optimisation of airport landside and airside processes.


9/22



3.4 Socioeconomic impact of ATM disruptions

In recent years, there have been numerous examples of severe storms and other natural events that have

caused widespread disruption to air transport, leaving passengers stranded in airports and cities all over the

world (ICAO, 2013). Flight disruptions are defined as situations where a scheduled flight is cancelled, or

delayed for two hours or more, within 48 hours of the original scheduled departure time (European Union,

2004). Disruptions impose capacity reductions and are measured in terms of traffic impact and delay impact

(EUROCONTROL, 2015). In the worst cases, the disruption extends well beyond the immediate geographical

region, causing a knock-on effect to passengers travelling to other regions.

The recent volcanic ash cloud crises have shown that a network approach is crucial to minimise the impact of

disruptions, and that a better understanding of disruptions and their effects is essential for the effective

design and implementation of a comprehensive set of measures at network level. The ash cloud crisis in

2010 triggered numerous analyses related to the impact of disruptions. More than 100,000 flights and 10million passengers were affected and 5,000 additional flights took place to reposition aircraft and crews and

to repatriate passengers. EUROCONTROL analysed how the 46% of cancelled flights and 43% of delayed

flights impacted on the future air traffic forecast due to the need to adjust historical (April and May 2010)

data (EUROCONTROL, 2010c). Other studies addressed the economic impact of the ash cloud consequences.

IATA focused on airlines losses, concluding that the closures of the European airspace caused airline losses of

around US$400 million per day from scheduled services (IATA, 2010). These numbers are the most

conservative estimation based on the verifiable data available on scheduled services. Starting from these

data, other analyses were conducted aiming to measure not only the economic impact of the disruption on

the airlines but also on alternative modes of transport, like car rentals and railways. As an example, Eurostar

reported that it carried 50,000 extra passengers on 15 April, and registered an increase of 33% on 17 April.P&O Ferries of France declared that their services between Britain, Spain, France and the Netherlands were

fully booked and that they had to employ extra personnel.

One reasonable hypothesis is that there are differences in passengers’ behaviour if the disruption can be

known in advance or not. However, the conventional analyses do not reveal such differences. For example,

the analyses of the Air France-KLM strike in September 2014 report a €400 million cost for the airlines, 8,500

flight cancellations and 1 million passengers affected, but there is little information about ticket changes or

other means of transport (Anna Aero, 2014; The New York Times, 2014).

Finally, disruptions have other consequences well beyond the direct economic impact for airlines (Rose,

2014): passengers might miss important business or personal events, and other economic sectors (e.g.,tourism) can be severely damaged. So far, the attempt to measure these indirect costs has had a moderate

success. Here again, the large amount of data generated by smart devices offers new opportunities to study

passengers’ behaviour in situations of disruption and measure how mobility patterns, travel times and

expenditure flows are modified.


10/22



4.

The BigData4ATM project

BigData4ATM (www.bigdata4atm-sesar.eu) is a research project within SESAR Exploratory Research which

investigates how different passenger-centric geolocated data can be analysed and combined with more

traditional demographic, economic and air transport databases to extract relevant information about

passengers’ behaviour, and how this information can be used to inform ATM decision making processes.

4.1 Project objectives

BigData4ATM pursues the following objectives:

1. to develop a set of methodologies and algorithms to acquire, integrate and analyse multiple

distributed sources of non-conventional ICT-based spatio-temporal data — including mobile phonerecords, data from geolocation technologies, credit card records and data from Internet social

networks, among others — with the aim of eliciting passengers’ behavioural patterns;

2. to develop new theoretical models translating these behavioural patterns into relevant and actionable

indicators for the planning and management of the ATM system;

3. to evaluate the potential applications of the new data sources, data analytics techniques and

theoretical models through a number of case studies relevant for the European ATM system, including

the development of passenger-centric door-to-door delay metrics, the improvement of air traffic

forecasting models, the analysis of intra-airport passenger behaviour and its impact on ATM, and the

assessment of the socioeconomic impact of ATM disruptions.

4.2 Approach

With the emergence of big data, some voices have raised concerns about the risk of focusing on descriptive

work and predictive, non-explanatory models, abandoning theory (Boyd, 2010; Graham, 2012).

BigData4ATM will adopt an integrative approach between empirical data analysis and theoretical modelling,

with the aim to take advantage of the opportunities offered by big data for the formulation, calibration, and

testing of new models of passengers’ behaviour and their interaction with the European ATM system.

The proposed research strategy comprises three main stages:

Data collection and assessment Big data analysis and modelling

Case studies

The first stage will deal with the collection of the different datasets required for the execution of the project

and the analysis of their strengths and limitations. This analysis will help refine the questions that will be

addressed during the second stage, aimed at extracting relevant behavioural information from the different

passenger-centric sources of georeferenced data. Finally, the third stage will evaluate the potential of these

new data sources to improve the quality of decision making in ATM in several specific applications.

http://www.bigdata4atm-sesar.eu/http://www.bigdata4atm-sesar.eu/


11/22



Data collection and assessment

Data collection

BigData4ATM has as one of its cornerstones the integration of massive repositories of heterogeneous data,

with special focus on new spatio-temporal data coming from smart technologies, but also including more

traditional data sources for calibration or validation purposes. The datasets that will be analysed include:

Twitter geolocated data collected using the Twitter streaming APIs. Twitter geolocated data contain

information about the location of the mobile device (latitude, longitude) each time a tweet is sent, as

long as geolocation is enabled. Potential uses of these data are the reconstruction of door-to-door

origin-destination matrices and passenger connectivity, and the semantic analysis of tweets to analyse

information propagation (e.g., efficiency of information broadcast about ATM disruptions) and social

perception of ATM;

anonymised mobile phone records, available through private commercial agreements with mobile

network operators. Anonymised mobile phone records (call detail records, CDRs) provide location

information about millions of users with high temporal and spatial resolution. CDRs are generated

when a mobile phone connected to the network makes or receives a phone call or an SMS, or

connects to the Internet. For invoicing purposes, the information regarding the base transceiver

station (BTS) tower to which the user was connected when the call or service was initiated and ended

is logged, providing an indication of the geographical position of the user. Mobile phone records

provide higher temporal resolution than mobile apps, as well as bigger and presumably less biased

samples, allowing the obtention of more accurate origin-destination matrices and the exploration of

problems like the connection of air transport with other modes. On the other hand, while data from

Twitter can be used almost worldwide, the usefulness of the data provided by a mobile network

operator is in most cases confined to the national level;

geolocation information from iBeacons and Wi-Fi hotspots in airports, which can provide valuable

information about passengers’ behaviour within the airport. These sensors detect mobile devices in

their surroundings, providing an estimation of the device location with a high level of accuracy; and

anonymised electronic transactions (credit and debit card usage records), which will be used to

analyse passenger expenditure patterns. Each transaction contains information about the amount

spent, business type (e.g., supermarket) and location (latitude, longitude), as well as some

sociodemographic information (e.g., age, gender, etc.) of the clients.

These new data sources will be complemented when necessary with more conventional data sources, whichwill serve a twofold purpose: (i) they will be useful for calibrating and validating the models built on the new

geolocation data sources; (ii) they will be blended with the new geolocation data to enrich the information

provided by each dataset separately. The data that will be used include:

public databases on sociodemographic data, data on economic activities, etc. available upon request

or accessible through open data initiatives;

air transport and ATM databases. Global origin-destination and passenger connectivity data, schedule

information, etc. will be obtained from aviation passenger intelligence solutions provided by

companies like Sabre and OAG. Data on air traffic delay are expected to be obtained from

EUROCONTROL’s Central Off ice for Delay Analysis (CODA).


12/22



Data quality assessment

The collected datasets will be assessed on validity, integrity, quality, and spatial and temporal resolution,

which will allow us to define in more detail the scope of the analyses that can be performed on each dataset(or each combination of datasets), as well as the associated limitations, leading to the final set of research

questions to be tackled in the data analysis and modelling stage.

Big data analysis and modelling

The previous datasets will be mined and analysed to extract relevant and actionable information. The goal is

to understand how traditional data sources (e.g., air traffic data) can be merged with new ICT-based

geolocation data to develop an enhanced understanding of passenger behavioural drivers and ATM

socioeconomic impact. The data analysis process will typically involve the following steps:

1.

pre-processing and data cleaning, to clean up errors and prepare the data structures;

2. study of the representativeness of the new digital data sources when used as a passengers’ survey

sample and correction of the sample to minimise bias;

3. knowledge extraction. To extract knowledge from the data, we will make use of a variety of

techniques including exploratory data analysis and visual analytics, spatial statistics and machine

learning methods. Classical approaches aimed at the analysis of spatial data from a static viewpoint

will be combined with techniques for the analysis of the temporal dynamics and the development of

explanatory and predictive models;

4. extrapolation to the total population.

The research questions that will be addressed can be classified in four main groups: (i) door-to-door mobilityanalysis, (ii) intra-airport passenger movement analysis, (iii) expenditure analysis and (iv) opinion and

sentiment analysis.

Door-to-door mobility analysis

The purpose here is to explore the potential of geolocated data from smart devices to capture passengers’

door-to-door mobility, from their real origin (e.g., their residence areas) to their final destination. Modal

origin-destination matrices and travel times in normal and abnormal conditions will be inferred from mobile

phone records within single countries and from Twitter for the whole continent, building on recent work in

this area (Andrienko et al., 2013; Lenormand et al., 2014; Picornell et al., 2015). The quality of the origin-

destination matrices will be assessed by comparing them with data from other sources, such as travelsurveys. Finally, we will develop different types of models of daily people flows and network evolution

(including spatial interaction models, reaction-diffusion models and models of complex evolving networks)

aimed to reproduce the observed patterns.

Intra-airport passenger movement analysis

Geolocation data from iBeacons and Wi-Fi hotspots in airports will be analysed to better characterise

passengers’ behaviour inside the airport and measure the time spent in the check -in desks, security control,

passport control and boarding queues. We will explore the potential of this information to predict

bottlenecks and to inform the formulation, calibration and validation of reliable models of passenger

movement and activity such those proposed by Schultz and Fricke (2011).


13/22



Expenditure analysis

The credit card dataset will be used to extract expenditure patterns within and outside airports, here again

building on recent work in the field (Lenormand, 2015a). We will explore the impact of different events onexpenditure flows, and will propose different types of spatial interaction models, such as Boltzmann, Lotka

and Volterra models (Wilson, 2010), trying to reproduce the coupling between short-term dynamics (daily

flows) and the structural evolution of the expenditure flows.

Opinion and sentiment analysis

Natural language processing, text analysis and computational linguistics will be used to analyse the content

of the public messages in Twitter (and possibly other social networks) to identify and extract subjective

information about air transport and ATM. Keyword spotting, lexical affinity and machine learning techniques

will be used to discriminate positive from negative opinions and tag emotional states. We will first explore

the public perception about ATM stakeholders in normal conditions, and will then analyse how it is modified

in the presence of disruptions (e.g., a strike), in search for the positive or negative impression of the public

towards different agents. The temporal evolution of these aftershocks will be analysed to explore whether

public attention vanishes in time as with other regular pieces of information (e.g., memes) or shows any

particularities such as a longer recovery processes or attention spans.

Case studies

The purpose of the case studies is to explore, evaluate and illustrate the potential of the insights on

passenger behaviour acquired from the new data sources to feed ATM decision-making processes. We have

identified four case studies where the bidirectional nature of the cause-effect relationships betweenpassengers’ behaviour and ATM performance is considered particularly relevant: (i) passenger-centric door-

to-door delay metrics, (ii) air traffic forecasting; (iii) integrated optimisation of airport landside and airside

processes, and (iv) socio-economic impact of ATM disruptions. Common to the four case studies is the idea

of exploring the opportunities offered by new passively collected data to find out what happened on any

particular day without having to plan in advance the collection of data for that day. This will allow us to learn

from “spontaneous experiments” and better understand how passenger behaviour changes as a result of

disruptive events, such as ash clouds or strikes, and how such behavioural changes feed back into the ATM

system and influence its performance. The specific questions to be addressed in each case study are defined

with a certain degree of flexibility, so that they can be refined by taking into account the results of the data

analysis work, as well as the inputs provided by relevant stakeholders.

Case study 1: passenger-centric door-to-door delay metrics

Case study 1 will focus on the estimation of passenger-centric door-to-door delay indicators and the

evaluation of ATM contribution to such delay. The SESAR WP-E POEM project has shown the importance of

passenger-centric metrics in fully assessing ATM system performance, and has quantified passenger value of

time as a function of delay at the final destination (Cook at al., 2013b). We will investigate how to extend this

work beyond the air segment of the trip, by combining information on air transport delays with the door-to-

door origin-destination pairs and travel times obtained from the new geolocation data, with the purpose of

providing new insights into the contribution of ATM to the European goal of at least 90% of travellers within

Europe able to complete a door-to-door journey within 4 hours.


14/22



Case study 2: air traffic forecasting

In this case study, new demand forecasting models will be developed. Traffic forecasts and peak-period

parameters are important for ATM systems planners to anticipate where and when congestion will occurand plan the implementation of CNS/ATM systems (ICAO, 2006). Currently, air traffic forecasts typically

combine flight statistics with econometric models that relate air traffic to observed variables whose

correlation with the traffic values is plausible, such as demographic, macroeconomic and tourism variables.

Our hypothesis here is that the information extracted from the new data sources can bring several

improvements in traffic forecasts, including:

larger, cheaper and permanently updated behavioural data samples for model validation and

calibration;

multimodal information for door-to-door itineraries, which will help study competition and synergies

between transport modes and quantify the impact of airport access/egress times on demand flows,enabling decision makers to understand the value of improved road or rail infrastructure;

identification of new behavioural variables. Mobility and expenditure patterns, as well as other

sociodemographic information inferred from the new ICT-based geolocated data sources, will be used

for travellers’ classification, identifying differences in sensitivity to economic and service changes and

route preferences. The newly developed door-to-door delay indicators will allow us to investigate the

relationship between demand and travel time reliability. Finally, we will explore the possibility of using

opinion and sentiment analysis to analyse the long-term cumulative effects of delays on demand.

Case study 3: integrated optimisation of airport landside and airside processes

Central to the vision for the European aviation sector by 2050 formulated in the Flightpath 2050 report aretime efficiency — meaning reduced door-to-door journey times, seamless inter-modal connections and

reliability — and resilience against disruptive events. The SESAR concept of operations will drive

improvements in the procedures being used by all ATM stakeholders, including airports. SESAR Project ID

06.05.04 D01 (SESAR, 2011) describes the way the AirPort Operations Centre (APOC) fits into the overall

framework of the SESAR operational concept. While the main APOC responsibilities are related with the

operational plan and the airside processes, these responsibilities also include “monitoring all closely linked

landside/terminal processes with direct influence on the Airport and the Network Operations Plan”. Recent

research has shown that passengers’ behaviour at the terminal and landside has a major impact o n flights

delays (EUROCONTROL, 2009a). The aim of this case study is to analyse the time spent by passengers in

airports looking for bottlenecks and improvement areas, with the ultimate goal of grasping the impact onthe ATM system. Geolocation data will be used to identify areas of congestion within the airport at given

times and their correlation with special events, airline schedules and other relevant factors. Tracking

algorithms will be used to characterise passenger movements inside the airport and estimate how much

time travellers spend in each step of their journeys, analysing the areas where the waiting time can be

reduced to achieve the 4-hours door-to-door objective. Queuing analysis will be used to estimate how much

time is spent in the check-in, security, passport control and boarding queues, and how this time can be

reduced (e.g., optimal number of passport/security control desks, etc.). Finally, we cannot forget that

airports are business areas, each time closer to shopping centres (The Independent, 2015; Retail Week,

2014). Reducing the time the passengers spend in the airport may therefore have an economic impact on

both passengers and business located in the airport. We will explore the possibility of using the credit carddataset to measure the impact of time efficiency improvements on passenger expenditures.


15/22



Case study 4: socioeconomic impact of ATM disruptions

Air travel disruptions (e.g., ash cloud, severe bad weather, controllers’ strike, etc.) alter passengers’ travel

plans, often leaving them trapped in airports or their surroundings for hours. The objective of this case studyis to explore how the information extracted from passenger-centric geolocated data can help assess the

impact of ATM disruptions in a more comprehensive manner, extending existing disruption analyses to

better take into account the direct and indirect effects on the passengers and society at large. For this

purpose, we will compare different socioeconomic and behavioural indicators during normal days and days

affected by major disruptions. Examples of the questions that will be investigated are the following:

the modification mobility patterns, travel times and expenditure patterns in the presence of ATM

disruptions;

the impact of disruptions on the accessibility of European regions;

the measurement of indirect economic costs of the disruptions by comparing visitors’ expenditure indifferent cities/regions;

passengers re-routing and transfer to alternative transport modes;

passengers’ opinions and perception of the ATM stakeholders.

Figure 1. BigData4ATM overall concept

Passenger-Centric geolocated dataDemographic and

socioeconomic

statistics

Door-to-door

mobility

Intra-airport

movement

Opinion and

sentiments…

Passenger-centric

door-to-door delay

indicators

Expenditure

patternsOther

Data analysis and modelling

Application to ATM decision-making processes

Integrated optimisation

of airport landside and

airside

Improved

traffic

forecasts

Socio-economic

impact of ATM

disruptions

Other

applications…

Air transport and

ATM databases


16/22



4.3 Target outcomes and expected impact

The project will deliver three main outcomes:

a set of novel methodologies, algorithms and tools for translating passenger-centric geolocated data

into meaningful behavioural information, such as door-to-door travel information, passenger

intra-airport movements, expenditure patterns, or customer satisfaction levels;

a set of new models and indicators built on passenger-centric data (e.g., accessibility of European

regions and door-to-door travel times) providing an improved comprehension of the interrelationships

between passengers’ behaviour, the ATM system and European society; and

a set of case studies demonstrating and evaluating the potential of the new data, models and

indicators to provide new insights into the planning and management of the ATM system.

These outcomes are expected to render a number of benefits for the European ATM sector:

Improvement of the quality and availability of knowledge for decision-making . Information about

passenger behaviour has so far been practically absent from ATM decision-making processes, largely due to

the difficulties to collect accurate and updated behavioural data. The pervasive penetration of smart mobile

devices opens the opportunity to fill this gap, by gathering rich data on citizens’ attitudes and activities. The

data sources investigated by BigData4ATM are expected to provide new insights about the relationships

between passenger behaviour and ATM performance. These insights will inform the development of new

theoretical models, metrics and indicators to translate the new datasets into actionable information for the

planning and management of the ATM system.

More agile ATM system designs. The integration of actionable passenger-centric information may

contribute to rendering the ATM system more resilient to challenges such as rapid changes in demand or

disruptive events. The ability to evaluate ATM performance from the point of view of their impact on

passengers and society at large, both in nominal conditions and in situations of performance degradation, is

expected to provide valuable inputs for the design of agile ATM concepts and systems.

Exchanging passengers/freight information seamlessly between different transport modes. One of the

results of BigData4ATM will be the extraction of modal split and travel time information for the different

segments of the door-to-door itineraries. This information will serve to study the complementarity and

competition between transport modes, as well as the interdependencies between airport connectivity and

passengers’ route choices. This knowledge will be of great value for improving the assessment of congestion

externalities and defining flight prioritisation criteria that take into account the full range of ATM

socioeconomic impacts. The results of these studies will contribute to identify relevant information that

might be integrated in future intermodal information systems.

Performance benefits across the entire ATM system. BigData4ATM will contribute to define new passenger-

centric metrics and indicators providing new angles of analysis of ATM performance, contributing to make

progress towards an ATM system where traffic management decisions are ultimately driven by the

passengers’ needs.


17/22



References

Airports Commission (2013), Discussion Paper 01: Aviation Demand Forecasting.

Andrienko, G., N. Andrienko, H. Bosch, T. Ertl, G. Fuchs, P. Jankowski, D. Thom (2013) Discovering Thematic

Patterns in Geo-Referenced Tweets through Space-Time Visual Analytics, Computing in Science and

Engineering, 2013, v.15(3), pp.72-82. http://doi.ieeecomputersociety.org/10.1109/MCSE.2013.70

Anna Aero (2014). Impact of Air France strike at French airports analysed; Paris CDG sees 12.3% drop in

September traffic. Available at http://www.anna.aero/2014/10/22/impact-of-air-france-strike-at-french-

airports-analysed/

Bagrow, J.P., Lin, Y.-R. (2012) Mesoscopic structure and social aspects of human mobility. PLoS ONE 7,

e37676.

Borge-Holthoefer, J., Rivero, A., Garcia, I., Cauhe, E., Ferrer, A. et al. (2011) Structural and dynamical patterns

on online social networks: The Spanish may 15th movement as a case study. PLoS ONE 6, e23883.

Boyd, D. (2010) Privacy and Publicity in the Context of Big Data. WWW 2010 conference, Raleigh, North

Carolina, 29 April 2010.

Brockmann, D., Hufnagel, L. and Geisel, T. (2006) The scaling laws of human travel. Nature 439, 462-465.

Cheng, L. (2014) Modelling Airport Passenger Group Dynamics Using an Agent-based Method, Masters by

Research thesis, Queensland University of Technology.

Conover, M.D., Davis, C., Ferrara, E., McKelvey, K., Menczer, F., et al. (2013) The geospatial characteristics of

a social movement communication network. PLoS ONE 8, e55957.

Cook, A. and Tanner, G (2005) “Citizens” Study - Results of European Focus Groups Examining Public

Perceptions of Air Transport Growth and ATM EEC/SEE/2005/013

Cook A., Tanner G., Cristóbal S. and Zanin M., (2012). Passenger-Oriented Enhanced Metrics, Proceedings of

the 2nd SESAR Innovation Days.

Cook A., Tanner G. and Zanin M., (2013a). Towards superior air transport performance metrics – imperativesand methods, Journal of Aerospace Operations, DOI 10.3233/AOP-130032.

http://dx.doi.org/10.3233/AOP-130032.

Cook A., Tanner G., Cristóbal S. and Zanin M., (2013b). New perspectives for air transport performance, in

Schaefer, D. (Ed.) Proceedings of the 3rd SESAR Innovation Days.

Crane, R. and Sornette, D. (2008). Robust dynamic classes revealed by measuring the response function of a

social system. PNAS 105, 15649-15653.

EUROCONTROL (2009a) Impact Study of Landside Elements on Airport Capacity and Delays.

EUROCONTROL (2009b) Challenges of Air Transport 2030 Survey of experts’ views.


18/22



EUROCONTROL (2010a) Air Transport Socio-Economic Studies at

http://www.eurocontrol.int/eec/public/standard_page/proj_Strategic_soc_eco_studies.html

EUROCONTROL (2010b) Strategic and Socio-Economic Studies at EUROCONTROL Synthesis 2010

EUROCONTROL (2010c) Ash-cloud of April and May 2010: Impact on Air Traffic.

EUROCONTROL/CND/STATFOR, June 2010.

EUROCONTROL (2015) Annual Network Operational Report 2014. EUROCONTROL June 2015.

European Commission (2010) Strategy for equality between women and men 2010-2015.

European Commission (2011a) White Paper: Roadmap to a Single European Transport Area – Towards a

competitive and resource efficient transport system. COM(2011) 144 final. Brussels, March 2011.

European Commission (2011b) Flightpath 2050. Europe’s Vision for Aviation. Report of the High Level Group

on Aviation Research.

European Commission (2013) Horizon 2020 Work Programme 2014-1015 - Smart, green and integrated

transport.

European Union (2004) Regulation (EC) No 261/2004 establishing common rules on compensation and

assistance to passengers in the event of denied boarding and of cancellation or long delay of flights.

Official Journal of the European Union, February 2004.

Frias-Martinez, V., Soto, V., Hohwald, H. and Frias-Martinez, E. (2012) Characterizing urban landscapes usinggeolocated tweets. In IEEE SocialCom/PASSAT pages 239-248.

Grabowicz, P.A., Ramasco, J.J., Moro, E., Pujol, J.M. and Eguíluz, V.M. (2012) Social Features of Online

Networks: The Strength of Intermediary Ties in Online Social Media. PLoS ONE 7, e29358.

Grabowicz, P.A., Ramasco, J.J., Gonçalves, B. and Eguiluz, V.M. (2014) Entangling Mobility and Interactions in

Social Media. PLoS ONE 9, e92196.

Gonçalves, B., Perra, N. and Vespignani, A. (2011) Validation of Dunbar’s number in Twitter conversations.

PLoS ONE 6, e22656.

Gonzalez-Bailon, M., Borge-Holthoefer J., Rivero, A. and Moreno, Y. (2011) The dynamics of protest

recruitment through an online network. Scientific Reports 1, 197.

Gonzalez, M.C., Hidalgo, C.A. and Barabasi, A.-L. (2008) Understanding individual human mobility patterns.

Nature 453, 779-782.

Graham, M. (2012) Big data and the end of theory? The Guardian, 9 March 2012.

Hawelka B., Sitko I., Beinat E., Sobolevsky S., Kazakopoulos P., et al. (2013) Geolocated twitter as a proxy for

global mobility patterns. ArXiv e-print arXiv:1311.0680.

IATA (2008) Air Travel Demand, Economics Briefing No 9.


19/22



IATA (2010) The impact of Eyjafjallajokull’s volcanic ash plume. IATA Economic Briefing, May 2010.

ICAO (2006) Doc 8991 AT/722/3 Manual on Air Traffic Forecasting.

ICAO (2013) Passengers Protection under Cases of Flights Disruptions. Worldwide Air Transport Conference

Sixth Meeting ICAO, Montreal, Canada, ACI 2013.

Jurdak, R. et al. (2015) Understanding human mobility from Twitter. http://arxiv.org/abs/1412.2154.

Kinchin B.(2004) A Synthesis of ATM Public Perception Surveys, EEC/SEE/2004/001

Lazer, D., Pentland, A., Adamic, L., Aral, S., Barabasi, A.-L., et al. (2009) Computational social science. Science

323, 721-723.

Lehmann, J., Ramasco, J.J., Gonçalves, B. and Catutto, C. (2012) Dynamical Classes of Collective Attention inTwitter. Procs. WWW, Lyon.

Lenormand, M., Picornell, M., Garcia Cantu, O., Tugores, A., Louail, T., Herranz, R., Barthelemy, M., Frias-

Martinez, E. and Ramasco, J.J. (2014) Cross-checking different source of mobility information. PLoS ONE

9, e105184.

Lenormand, M., Tugores, A., Colet, P. and Ramasco, J.J. (2014b) Tweets on the road. PLoS ONE 9, e105407.

Lenormand, M., Picornell, M., Garcia Cantu, O., Tugores, A., Louail, T., Herranz, R., Barthelemy, M., Frias-

Martinez, E, San Miguel, M. and Ramasco, J.J. (2015a) Comparing and modeling land use organization in

cities. http://arxiv.org/abs/1503.06152.

Lenormand, M., Tugores, A., Gonçalves, B. and Ramasco, J.J. (2015b) Human diffusion and city influence.

http://arxiv.org/abs/1501.07788.

Lenormand, M., Louail, T., Garcia Cantu, O., Picornell, M., Herranz, R., Murillo Arias, J., Barthelemy, M., San

Miguel, M. and Ramasco, J.J. (2015c) Influence of sociodemographic characteristics on human mobility.

Scientific Reports (In press),

Liben-Nowell D., Novak J., Kumar R., Raghavan P., Tomkins A. (2005) Geographic routing in social networks.

PNAS 102, 11623-11628.

Louail, T., Lenormand, M., Garcia-Cantu, O., Picornell, M., Herranz, R., Frias-Martinez, E., Ramasco, J.J. and

Barthelemy, M. (2014) From mobile phone data to the spatial structure of cities. Scientific Reports 4,

5276 (2014).

Louail, T., Lenormand, M., Garcia-Cantu, O., Picornell, M., Herranz, R., Frias-Martinez, E., Ramasco, J.J. and

Barthelemy, M. (2015) Uncovering the spatial structure of mobility networks. Nature Communications 6,

6007.

Mahaud, P. and Arrighi de Casanova, D. (2004) What Image of ATM? An Analysis of 2002-2003 European

Press EEC/SEE/2004/002

Mocanu, D., Baronchelli, A., Perra, N., Gonçalves, B. and Vespignani, A. (2013) The Twitter of Babel: Mapping

World Languages through Microblogging Platforms, PLoS One 8, E61981.


20/22



Noulas, A., Scellato, S., Lambiotte, R., Pontil, M. and Mascolo, C. (2012) A tale of many cities: Universal

patterns in human urban mobility. PloS one 7, e37027.

OECD ITF (2014) Contemporary Airport Demand Forecasting: Choice Models and Air Transport Forecasting,

Discussion Paper 2014-7.

Onnela J.-P., Saramaki J., Hyvonen J., Szabo G., Lazer D., et al. (2007) Structure and tie strengths in mobile

communication networks. Proc. Natl. Acad. Sci. U.S.A. 104, 7332-7336.

Onnela, J.-P. and Reed-Tsochas, F. (2009) Spontaneous emergence of social influence in online systems.

PNAS 107, 18375-18380.

Picornell M., T. Ruiz, M. Lenormand, J.J. Ramasco, T. Dubernet and E. Frías-Martínez (2015). Exploring the

potential of phone call data to characterize the relationship between social network and travel behavior.

Transportation, DOI: 10.1007/s11116-015-9594-1

Retail Week (2014) Analysis: How travel retail went from waste of time to strategically crucial -at:

http://www.retail-week.com/property/analysis-how-travel-retail-went-from-waste-of-time-to-

strategically-crucial/5058704.article.

Rose, N (2014). Passengers First, Re-thinking irregular operations. Amadeus IT Group.

Schultz, M. and Fricke, H. (2011) Managing Passenger Handling at Airport Terminals: Individual-based

Approach for Modeling the Stochastic Passenger Behavior. 9th USA/Europe ATM Research and

Development Seminar (ATM2011), Berlin, Germany.

SESAR (2007) Definition Phase Deliverable 3: The ATM Target Concept DLM-0612-001-02-00

SESAR (2011) Project ID 06.05.04 AirPort Operations Centre (APOC) definition, D01 Initial Operational

Concept.

Song C., Qu Z., Blumm N., Barabasi A.-L. (2010) Limits of predictability in human mobility, Science 327, 1018.

Soto, V. and E. Frías-Martinez (2011) Automated land use identification using cell-phone records. Procs. of

the ACM conference HotPlanet'11 pp. 17-22, New York.

The Independent (2015), Cover Story: Have airports become shopping centres with runways attached? -at:http://www.independent.co.uk/news/cover-story-have-airports-become-shopping-centres-with-

runways-attached-1344426.html

The New York Times (2014). Air France Puts Cost of Pilots’ Strike at More Than $400 Million Available at

http://www.nytimes.com/2014/10/09/business/international/air-france-puts-cost-of-pilots-strike-at-

more-than-400-million.html

Tizzoni, M. et al. (2014) On the use of human mobility proxy for the modeling of epidemics. PLoS

Computational Biology 10, e1003716.

Wilson, A. (2010) Entropy in Urban and Regional Modelling: Retrospect and Prospect. Geographical Analysis42 (2010) 364 –394.


21/22

This project has received funding from the SESAR Joint Undertaking under

grant agreement No 699260 under European Union’s Horizon 2020research and innovation programme. Opinions expressed in this work

reflect only the authors’ view; the SJU shall not be considered responsible

for any use that may be made of the information contained herein.


22/22

www.access-project.eu

www.bigdata4atm-sesar.eu
http://www.access-project.eu/http://www.access-project.eu/

bigdata4atm whitepaper - may 2016

Documents